Suyash Dombe

I’m a results-driven Data Professional focused on AI evaluation, RLHF/annotation QA, and training-data quality. In my current role as an AI Speech Evaluation Specialist (TTS Quality Analyst), I assess 300+ Hindi/Marathi/English TTS audio samples and score Pronunciation Faithfulness, Naturalness, Intonation, Pacing, and Speaker Persona Similarity using detailed rubrics.

I also build systems and workflows that translate evaluation into measurable improvements—authored structured written rationales that directly support multilingual speech model training data, and run comparative human evaluations to guide model ranking and selection. Beyond speech, I’ve delivered analytics-ready datasets by cleaning 20,000+ records in Advanced Excel (VBA, pivot tables) and using AI/LLM extraction, engineered Python scraping pipelines for 35,000+ contacts, analyzed 500,000+ retail transactions with Power BI, and developed full end-to-end RAG pipelines (LangChain + FAISS + Whisper + local LLaMA inference) for production-like learning experiences.